Chronic hepatitis B virus (HBV) infection represents a serious public health problem and is one of the major causes of hepatocellular carcinoma (HCC) worldwide. Among patients with chronic hepatitis B (CHB), the risk of HCC development increases with the severity of cirrhosis (1). The annual HCC incidence for patients with CHB without cirrhosis is ~0.01–1.4%, while the HCC incidence increases to 0.9–10.8% in patients with cirrhosis (2, 3). With the introduction of nucleos(t)ide analog (NA) therapy over the past two decades, long-term viral suppression has emerged as the most dominant modifier of HCC in patients with CHB (4). In patients with cirrhosis, NA therapy with entecavir (ETV) or tenofovir disoproxil fumarate (TDF) results in a significant reduction (~30%) in HCC risk (3). However, HCC risk cannot be completely eliminated in patients receiving NA therapy, especially in patients with cirrhosis (5, 6). The high risk of HCC highlights the need for disease management in this at-risk population.

Early detection of HCC by periodic surveillance is critical for improving patient outcomes. HCC surveillance is currently recommended in all patients with cirrhosis; however, the risk of HCC varies widely among patients. Thus, an individual HCC risk prediction is of importance for implementing feasible and effective HCC screening. Therefore, numerous HCC risk scores have been developed. Early scores are developed in untreated patients with CHB and include factors such as age, sex, HBV DNA, core promoter mutations, and the cirrhosis-HCC score (GAG-HCC) (7), the Chinese University of Hong Kong-HCC score (CU-HCC) (8), risk estimation for HCC in CHB score (REACH-B) (9), liver stiffness measurement-HCC score (LSM-HCC) (10), and real-world score for HCC (RWS-HCC) (11). However, such scores may have led to an overestimation of HCC incidence in patients undergoing NA therapy. Therefore, several scores have been developed in patients with CHB undergoing NA therapy, including the modified REACH-B score (mREACH-B) (12), platelets, age, and gender score (PAGE-B) (13), cirrhosis, age, male sex, and diabetes mellitus score (CAMD) (14), and the modified PAGE-B score (mPAGE-B) (15). Recently, a novel scoring system that includes age, male sex, albumin-bilirubin, and platelets (aMAP) has been proposed, and its performance in assessing the 5-year HCC risk has been validated in several cohorts with different etiologies and ethnicities (16), but its performance for late HCC after a longer period of treatment warrants further research.

As various HCC risk scores are available and patients with CHB are a heterogeneous group, it is not easy for clinicians to determine if these scores can be applied in their clinical practice effectively and which HCC risk score should be used for specific patients. Moreover, the potential clinical utility of these HCC risk scores in patients with cirrhosis remains unclear, as patients with cirrhosis are excluded from many studies, and there has been no study focusing on this at-risk population (17). In this study, we externally validated the predictability of the aMAP score using an independent cohort of treatment-naïve patients with CHB-related compensated cirrhosis who were receiving ETV or TDF at different follow-up timepoints, especially outside its original intended time horizon. We compared the performance of the aMAP score with that of other risk scores. Furthermore, we incorporated other emerging risk factors to the aMAP score to enhance its predictive ability and clinical relevance. Moreover, we explored whether its predictability would change with re-assessment of the score during therapy.

The study protocol was conducted in accordance with the ethical guidelines of the Declaration of Helsinki 1975 and was approved by the Research Ethics Committee of Ruijin Hospital (No.2019-202). As this was a retrospective analysis of de-identified medical records, patient informed consent was not obtained.

Although the effect of antiviral therapy on reducing the risk of HCC development has been well-established, the substantial residual risk of HCC during long-term NA therapy may still exists, particularly in patients with cirrhosis (20). It is critical to determine a more accurate HCC prediction model optimized for disease management in at-risk patients with CHB-related cirrhosis. In this independent cohort study, we validated that the aMAP score has a superior or comparable predictive performance (Harrell's c-index: 0.724; 95% CI: 0.701–0.747) in predicting HCC development among patients with cirrhosis undergoing long-term ETV or TDF therapy. The large sample size and mid-long-term follow-up in this study enhance the statistical reliability of the results. Moreover, a sufficient number of patients developed HCC during follow-up, affording adequate statistical power. The relative homogeneity of our study population, well-defined population of CHB with compensated cirrhosis receiving ETV or TDF therapy, reduced potential confounding factors.

Older age, male sex, and lower platelet count are well-known risk factors associated with HCC development (13, 15, 16). In addition, higher serum total bilirubin levels are an independent risk factor for HCC development in this study, which is consistent with previous studies (8, 16). In this study, lower serum albumin levels were not found to be an independent risk factor for HCC development. This is most likely due to the fact that patients with decompensated or advanced cirrhosis (Child-Pugh class B or C) were excluded from our study.

Liver cirrhosis has been reported to be a risk factor for HCC development in patients with CHB with or without NA therapy (7, 14). In the present study, all patients had cirrhosis at enrollment based on liver biopsies or clinical, radiographic, and/or laboratory data. A clinical diagnosis of cirrhosis is more likely in real-world practice. Furthermore, the fact that cirrhosis is not included in the aMAP score is likely advantageous as long-term NA therapy can lead to the regression of histological cirrhosis in patients with CHB (21). The significantly improved predictive performance of the aMAP score after 1 year of NA therapy compared to its calculated at baseline may be explained by the specific variables included in the aMAP score. In other words, NA therapy as a risk modifier of HCC may have a salutary effect on the aMAP score as well. However, cirrhotic patients with a decline in aMAP from baseline to after 1 year of treatment did not exhibit a lower HCC risk than did those with an increased aMAP in our study. Consequently, a single short-term on-treatment change of aMAP was insufficient to determine the risk of HCC. As for the issue of whether the HCC risk was only apparent when aMAP scores moved between categories, future studies that combine aMAP and ΔaMAP will be better positioned to address it. Notably, our study validated that the aMAP score could predict HCC risk not only in the first 5 years of therapy, but also after 5 years of therapy. Since the aMAP score is easily computed using data from routine laboratory tests, it can be monitored regularly in patients undergoing NA therapy and meets the predictive needs of clinical practice.

The predictive performance of the aMAP score was comparable to that of the CAMD and mPAGE-B scores models and better than that of the PAGE-B score. The aMAP score may be useful for identifying a subgroup of patients with cirrhosis with a low risk of the development of HCC (aMAP <50 indicates a 2.9% risk of HCC within 5 years). Typically, surveillance strategies to detect early-stage HCC are cost-effective when the annual risk exceeds 0.2% in patients without cirrhosis and 1.5% in patients with cirrhosis. Therefore, the low-risk subgroup (which accounts for 14% of the study population) with an HCC incidence of <1.5%/year indicates the need to recalculate the cost-effectiveness of the current HCC surveillance practices. Patients with an aMAP score <50 achieved an NPV of 96.6% for the 5-year development of HCC, which is similar to the reported NPV of patients with cirrhosis when the PAGE-B related scores are used (22). When a more precise definition of low HCC risk is created by combining the aMAP and mPAGE-B scores, 11.6% of patients with cirrhosis undergoing NA therapy were classified as low risk with NPVs of 100 and 98.2% to exclude HCC at 3 and 5 years, respectively. Therefore, life-long HCC surveillance is recommended for patients with cirrhosis. Future research efforts will focus on exploring additional biomarkers to further identify patients with zero risk of HCC development for whom HCC surveillance may no longer be needed or may be able to be performed at longer intervals.

HCC risk scoring systems are also useful to identify patients with a high risk of HCC development and a potential need for intensified HCC surveillance. HCC surveillance is currently based on biannual abdominal ultrasounds for all patients with cirrhosis, regardless of the patient's actual HCC risk. This approach often results in an over- or underestimation of HCC risk for each patient (23). Additionally, the diagnostic accuracy of ultrasound is suboptimal in patients with cirrhosis, particularly for the detection of early-stage HCC. Dynamic CT or MRI offer higher sensitivity and specificity in detecting HCC, but have several limitations for general use in HCC surveillance (24). Therefore, patients with cirrhosis who are classified in the high-risk group may be candidates for specific HCC surveillance with more sensitive imaging modalities such as MRI (25). HCC risk scores offer clinically useful information by identifying subgroups of patients with CHB with an annual HCC risk exceeding 3–5% (17). In the present study, 37.3% of patients with cirrhosis undergoing NA therapy were classified in the high-risk group and among these patients 27.9% developed HCC at year 5. Like existing HCC risk scores, the PPV of the aMAP score at a cut-off point of 60 was not optimal. Moreover, the use of the “AND” and “OR” approaches with aMAP and other HCC scoring systems was not able to reach optimal PPV.

DM or prediabetes, as an important component of metabolic syndrome, has been found to be associated with an increased incidence of HCC (26, 27). CHB Patients with newly diagnosis of DM have an increased risk of cirrhosis and hepatic decompensation over time (28). DM is a risk factor for HCC development in patients with CHB and cirrhosis who undergo NA therapy (29). The association between DM and HCC risk has been reported to be independent of cirrhosis, although most patients with HCC presented with cirrhosis (30). In fact, it seems that DM is a bigger independent risk factor for HCC than any of the other variables examined in this study. Interestingly, further analysis of combination of DM and aMAP found that DM could help to stratify the aMAP high-risk group, but not the low- or intermediate-risk groups. The statistically inconsistent effect across aMAP risk groups might actually be attributed to confounding by other underlying variables (e.g., metabolic liver disease), that may be hard to determine within the limits of our dataset. It should also be noted that there is a large disparity in the proportions of diabetic patients among different aMAP risk groups in our study, 2.1, 5.3, and 14.0% in low-, intermediate-, and high-risk groups, respectively.

The biological mechanism for how DM influences HCC development is not well-understood, but several possibilities have been hypothesized. Insulin resistance and hyperinsulinemia, hallmarks of type 2 diabetes, are believed to play important roles in hepatocarcinogenesis. DM is also an important risk factor for the development of steatohepatitis, which may be associated with an increased risk of HCC. As DM is easy to assess in clinical practice and it significantly improves the predictive value of the aMAP score, we recommend routine screening for DM in patients with CHB with or without cirrhosis before the initiation of NA therapy. Metformin, which is administered to improve insulin sensitivity, has been reported to be associated with decreased HCC risk (31). Therefore, the proper management of DM and cautious selection of therapy for DM in patients with CHB, especially for those with a high risk of cirrhosis, are recommended.

This study is not without limitations. First, the present study was conducted solely based on a retrospective, single-center cohort in a large tertiary referral university hospital, and its results are potentially subject to selection bias. Second, although we validated the predictive performance of the initial aMAP score and the 1-year aMAP score, the predictive role of the combination of aMAP and on-treatment changes in aMAP (ΔaMAP) during NA therapy and their optimal time point requires further investigation, as in the case of the fibrosis index based on four factors (FIB-4) (32). Third, given that metabolic risk factors, including diabetes, obesity, high blood pressure, and hypercholesterolemia, are associated with an increased HCC risk in patients with CHB (33), other metabolic risk factors should be incorporated in future studies as such models will have higher discriminatory power. In addition, future research to determine how the control of glycemia, blood pressure, and lipids and/or the types of antidiabetics, antihypertensive, and lipid-lowering medications may affect the risk of HCC is required. Finally, the severity of liver fibrosis/cirrhosis and the longitudinal changes during NA therapy were not evaluated by transient elastography (TE) in this study as this technology was not available for the majority of the study period in our hospital. In addition, it is difficult to perform paired TE examinations in a large cohort of patients with CHB before and after long-term antiviral treatment in a real-life setting. The implementation of TE will likely improve the predictive performance of the aMAP score, similar to its effect on the LSM-HCC (10) and mREACH-B (12) scores.

In conclusion, the aMAP score achieved an acceptable and comparable HCC predictive performance in at-risk patients with compensated cirrhosis undergoing ETV or TDF therapy, not only calculated at baseline, but after 1 year of antiviral therapy as well. The addition of DM as a constituent to the aMAP score contributed to a marginal improvement in the prediction of extremely high HCC risk. These patients are suggested to undergo more intensive and effective HCC surveillance program for early diagnosis and to improve prognosis. In addition, these patients are optimal candidates for HCC chemopreventive agents, including metformin (31), interferon (34), and other therapies that may be developed.

The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

The studies involving human participants were reviewed and approved by the Research Ethics Committee of Ruijin Hospital. Written informed consent for participation was not required for this study in accordance with the national legislation and the institutional requirements.

QX and HG conceived and designed the study. HG, YH, GZ, LC, WC, and HW collected the clinical data. HG, YH, and QG conducted the analysis and prepared the manuscript. QG and QX critically revised the manuscript. All authors approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.